AI Agents

Agents that do the work, not just the talking

Most AI tools stop at answering questions. RootCX agents update records, chase approvals, and follow up with customers, under the same permissions as your team. Every action is logged.

What they can do

Not chatbots. Workers.

RootCX agents share the same database, users, and permissions as your internal apps. No plugins to wire up, no separate vector store to maintain.

Build

From idea to live agent, fast

Describe what the agent should do. RootCX generates the code, wires it to your data model, and gives you a one-click deploy. Review everything before it goes live.

Know

Grounded in your real data

Agents read from the same database your team uses every day. Answers reflect what is actually in your system right now, not a stale export from last month.

Act

Safe by default

Every tool call goes through RBAC, and every result is written to an immutable audit log. You see the caller, the action, and the outcome. Nothing happens behind your back.

Deep capabilities

Built-in, not bolted on

Every agent gets a full toolbox out of the box. No third-party plugins, no glue code, no fragile prompt chains.

Tools

Read, write, integrate, browse

Agents ship with typed tools for reading data, writing records, calling integrations, and browser automation. Each tool respects the agent's permissions.

query_datamutate_datacall_integrationinvoke_agentbrowser
Supervision

You choose how much trust to give

Three modes: autonomous (runs freely), supervised (proposes, you approve), strict (every action needs sign-off). Set rate limits and approval flows per tool.

Memory

Conversations that persist

Long threads are automatically compacted so the agent keeps context without blowing through token limits. Session history is stored per-user, not shared globally.

Delegation

Agents that call other agents

A triage agent can route to specialists. A research agent can hand off to a writer. Fleet-wide intelligence, not siloed bots.

Channels

Meet your team where they already talk

Connect an agent to Telegram, Slack, Discord, or any channel your team uses. Pick a bot, link it to an agent, and people start chatting with it. No new app to install, no training needed.

Security

Enterprise-grade agent security

Every agent action is governed by the same security model that protects your data and users.

See the CISO checklist for AI agents in production →
SSO via OIDC

Okta, Microsoft Entra ID, Google Workspace, Auth0. Your identity provider stays the source of truth.

Namespaced RBAC

Agents get assigned roles just like human users. Permissions checked on every tool call, not just at login.

Immutable audit log

Every agent action written to an append-only log at the trigger level. Who, what, when. Cannot be edited or deleted.

Encrypted vault

API keys stored AES-256 encrypted, injected at runtime. Never in agent code, never in logs.

Self-hostable

Run everything on your own servers. No data ever leaves your network. Same product, no feature gates.

Source-available

FSL-1.1-ALv2 license. Read the code, audit it against your requirements, run it wherever compliance needs it.

GET STARTED

Deploy your first agent today

Open-source and self-hostable. Ready for your security review before the conversation starts.